drm/xe: Add performance tunings to debugfs

Add a list of active tunings to debugfs, analogous to the existing
list of workarounds.

Rationale being that it seems to make sense to either have both or none.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Cc: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250227101304.46660-6-tvrtko.ursulin@igalia.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

drivers/gpu/drm/xe/xe_gt.c

@@ -362,6 +362,10 @@ int xe_gt_init_early(struct xe_gt *gt)
 	if (err)
 		return err;
 
+	err = xe_tuning_init(gt);
+	if (err)
+		return err;
+
 	xe_wa_process_oob(gt);
 
 	xe_force_wake_init_gt(gt, gt_to_fw(gt));

drivers/gpu/drm/xe/xe_gt_debugfs.c

@@ -30,6 +30,7 @@
 #include "xe_reg_sr.h"
 #include "xe_reg_whitelist.h"
 #include "xe_sriov.h"
+#include "xe_tuning.h"
 #include "xe_uc_debugfs.h"
 #include "xe_wa.h"
 
@@ -217,6 +218,15 @@ static int workarounds(struct xe_gt *gt, struct drm_printer *p)
 	return 0;
 }
 
+static int tunings(struct xe_gt *gt, struct drm_printer *p)
+{
+	xe_pm_runtime_get(gt_to_xe(gt));
+	xe_tuning_dump(gt, p);
+	xe_pm_runtime_put(gt_to_xe(gt));
+
+	return 0;
+}
+
 static int pat(struct xe_gt *gt, struct drm_printer *p)
 {
 	xe_pm_runtime_get(gt_to_xe(gt));
@@ -300,6 +310,7 @@ static const struct drm_info_list debugfs_list[] = {
 	{"powergate_info", .show = xe_gt_debugfs_simple_show, .data = powergate_info},
 	{"register-save-restore", .show = xe_gt_debugfs_simple_show, .data = register_save_restore},
 	{"workarounds", .show = xe_gt_debugfs_simple_show, .data = workarounds},
+	{"tunings", .show = xe_gt_debugfs_simple_show, .data = tunings},
 	{"pat", .show = xe_gt_debugfs_simple_show, .data = pat},
 	{"mocs", .show = xe_gt_debugfs_simple_show, .data = mocs},
 	{"default_lrc_rcs", .show = xe_gt_debugfs_simple_show, .data = rcs_default_lrc},

drivers/gpu/drm/xe/xe_gt_types.h

@@ -413,6 +413,16 @@ struct xe_gt {
 		bool oob_initialized;
 	} wa_active;
 
+	/** @tuning_active: keep track of active tunings */
+	struct {
+		/** @tuning_active.gt: bitmap with active GT tunings */
+		unsigned long *gt;
+		/** @tuning_active.engine: bitmap with active engine tunings */
+		unsigned long *engine;
+		/** @tuning_active.lrc: bitmap with active LRC tunings */
+		unsigned long *lrc;
+	} tuning_active;
+
 	/** @user_engines: engines present in GT and available to userspace */
 	struct {
 		/**

drivers/gpu/drm/xe/xe_tuning.c

@@ -7,6 +7,8 @@
 
 #include <kunit/visibility.h>
 
+#include <drm/drm_managed.h>
+
 #include "regs/xe_gt_regs.h"
 #include "xe_gt_types.h"
 #include "xe_platform_types.h"
@@ -140,10 +142,44 @@ static const struct xe_rtp_entry_sr lrc_tunings[] = {
 	{}
 };
 
+/**
+ * xe_tuning_init - initialize gt with tunings bookkeeping
+ * @gt: GT instance to initialize
+ *
+ * Returns 0 for success, negative error code otherwise.
+ */
+int xe_tuning_init(struct xe_gt *gt)
+{
+	struct xe_device *xe = gt_to_xe(gt);
+	size_t n_lrc, n_engine, n_gt, total;
+	unsigned long *p;
+
+	n_gt = BITS_TO_LONGS(ARRAY_SIZE(gt_tunings));
+	n_engine = BITS_TO_LONGS(ARRAY_SIZE(engine_tunings));
+	n_lrc = BITS_TO_LONGS(ARRAY_SIZE(lrc_tunings));
+	total = n_gt + n_engine + n_lrc;
+
+	p = drmm_kzalloc(&xe->drm, sizeof(*p) * total, GFP_KERNEL);
+	if (!p)
+		return -ENOMEM;
+
+	gt->tuning_active.gt = p;
+	p += n_gt;
+	gt->tuning_active.engine = p;
+	p += n_engine;
+	gt->tuning_active.lrc = p;
+
+	return 0;
+}
+ALLOW_ERROR_INJECTION(xe_tuning_init, ERRNO); /* See xe_pci_probe() */
+
 void xe_tuning_process_gt(struct xe_gt *gt)
 {
 	struct xe_rtp_process_ctx ctx = XE_RTP_PROCESS_CTX_INITIALIZER(gt);
 
+	xe_rtp_process_ctx_enable_active_tracking(&ctx,
+						  gt->tuning_active.gt,
+						  ARRAY_SIZE(gt_tunings));
 	xe_rtp_process_to_sr(&ctx, gt_tunings, &gt->reg_sr);
 }
 EXPORT_SYMBOL_IF_KUNIT(xe_tuning_process_gt);
@@ -152,6 +188,9 @@ void xe_tuning_process_engine(struct xe_hw_engine *hwe)
 {
 	struct xe_rtp_process_ctx ctx = XE_RTP_PROCESS_CTX_INITIALIZER(hwe);
 
+	xe_rtp_process_ctx_enable_active_tracking(&ctx,
+						  hwe->gt->tuning_active.engine,
+						  ARRAY_SIZE(engine_tunings));
 	xe_rtp_process_to_sr(&ctx, engine_tunings, &hwe->reg_sr);
 }
 EXPORT_SYMBOL_IF_KUNIT(xe_tuning_process_engine);
@@ -168,5 +207,25 @@ void xe_tuning_process_lrc(struct xe_hw_engine *hwe)
 {
 	struct xe_rtp_process_ctx ctx = XE_RTP_PROCESS_CTX_INITIALIZER(hwe);
 
+	xe_rtp_process_ctx_enable_active_tracking(&ctx,
+						  hwe->gt->tuning_active.lrc,
+						  ARRAY_SIZE(lrc_tunings));
 	xe_rtp_process_to_sr(&ctx, lrc_tunings, &hwe->reg_lrc);
 }
+
+void xe_tuning_dump(struct xe_gt *gt, struct drm_printer *p)
+{
+	size_t idx;
+
+	drm_printf(p, "GT Tunings\n");
+	for_each_set_bit(idx, gt->tuning_active.gt, ARRAY_SIZE(gt_tunings))
+		drm_printf_indent(p, 1, "%s\n", gt_tunings[idx].name);
+
+	drm_printf(p, "\nEngine Tunings\n");
+	for_each_set_bit(idx, gt->tuning_active.engine, ARRAY_SIZE(engine_tunings))
+		drm_printf_indent(p, 1, "%s\n", engine_tunings[idx].name);
+
+	drm_printf(p, "\nLRC Tunings\n");
+	for_each_set_bit(idx, gt->tuning_active.lrc, ARRAY_SIZE(lrc_tunings))
+		drm_printf_indent(p, 1, "%s\n", lrc_tunings[idx].name);
+}

drivers/gpu/drm/xe/xe_tuning.h

@@ -6,11 +6,14 @@
 #ifndef _XE_TUNING_
 #define _XE_TUNING_
 
+struct drm_printer;
 struct xe_gt;
 struct xe_hw_engine;
 
+int xe_tuning_init(struct xe_gt *gt);
 void xe_tuning_process_gt(struct xe_gt *gt);
 void xe_tuning_process_engine(struct xe_hw_engine *hwe);
 void xe_tuning_process_lrc(struct xe_hw_engine *hwe);
+void xe_tuning_dump(struct xe_gt *gt, struct drm_printer *p);
 
 #endif